update hicpro2higlass

bin/hicpro2higlass.sh

@@ -103,6 +103,7 @@ function help {
     echo "   -i|--input INPUT : allValidPairs or matrix file generated by HiC-Pro"
     echo "   -r|--res RESOLUTION : .matrix file resolution or maximum resolution to reach from the .allValidPairs input file"
     echo "   -c|--chrom CHROMSIZE : chromosome size file"
+    echo "   -p|--proc NB_CPU : number of CPUs for cooler"
     echo "   [-n|--norm] : run cooler matrix balancing algorithm"
     echo "   [-o|--out] : output path. Default is current path"
     echo "   [-t|--temp] TEMP : path to tmp folder. Default is current path"
@@ -125,6 +126,7 @@ for arg in "$@"; do
       "--bed")   set -- "$@" "-b" ;;
       "--res")   set -- "$@" "-r" ;;
       "--chrom") set -- "$@" "-c" ;;
+      "--proc") set -- "$@" "-p" ;;
       "--out") set -- "$@" "-o" ;;
       "--temp") set -- "$@" "-t" ;;
       "--norm")   set -- "$@" "-n" ;;
@@ -136,18 +138,20 @@ done
 INPUT_HICPRO=""
 INPUT_BED=""
 NORMALIZE=0
+NPROC=1
 CHROMSIZES_FILE=""
 RES=10000
 OUT="./"
 TEMP="./"
 
-while getopts ":i:b:c:r:o:t:nh" OPT
+while getopts ":i:b:c:p:r:o:t:nh" OPT
 do
     case $OPT in
 	i) INPUT_HICPRO=$OPTARG;;
 	b) INPUT_BED=$OPTARG;;
 	n) NORMALIZE=1;;
 	c) CHROMSIZES_FILE=$OPTARG;;
+	p) NPROC=$OPTARG;;
 	r) RES=$OPTARG;;
 	o) OUT=$OPTARG;;
 	t) TEMP=$OPTARG;;
@@ -223,9 +227,9 @@ if [[ $DATATYPE == "MATRIX" ]]; then
 
     echo -e "\nZoomify .cool file ..."
     if [[ $NORMALIZE == 1 ]]; then
-	cooler zoomify --balance $tmp_dir/$out
+	cooler zoomify --nproc ${NPROC} --balance $tmp_dir/$out
     else
-	cooler zoomify --no-balance $tmp_dir/$out
+	cooler zoomify --nproc ${NPROC} $tmp_dir/$out
     fi
     out=$(basename $INPUT_HICPRO | sed -e 's/.mat.*/.mcool/')
     
@@ -233,19 +237,19 @@ elif [[ $DATATYPE == "VALID" ]]; then
     out=$(basename $INPUT_HICPRO | sed -e 's/.allValidPairs.*/.cool/')
 
     awk '{OFS="\t";print $2,$3,$4,$5,$6,$7,1}' $INPUT_HICPRO | sed -e 's/+/1/g' -e 's/-/16/g' > $tmp_dir/contacts.txt
-    cooler csort --nproc 2 -c1 1 -p1 2 -s1 3 -c2 4 -p2 5 -s2 6 \
+    cooler csort --nproc ${NPROC} -c1 1 -p1 2 -s1 3 -c2 4 -p2 5 -s2 6 \
 	   -o $tmp_dir/contacts.sorted.txt.gz  \
 	   $tmp_dir/contacts.txt \
 	   $CHROMSIZES_FILE
     
     cooler makebins $CHROMSIZES_FILE $RES > $tmp_dir/bins.bed
-    cooler cload pairix $tmp_dir/bins.bed $tmp_dir/contacts.sorted.txt.gz $tmp_dir/$out
+    cooler cload pairix --nproc ${NPROC} $tmp_dir/bins.bed $tmp_dir/contacts.sorted.txt.gz $tmp_dir/$out
 
     echo -e "\nZoomify .cool file ..."
     if [[ $NORMALIZE == 1 ]]; then
-	cooler zoomify --balance $tmp_dir/$out
+	cooler zoomify --nproc ${NPROC} --balance $tmp_dir/$out
     else
-	cooler zoomify --no-balance $tmp_dir/$out
+	cooler zoomify --nproc ${NPROC} $tmp_dir/$out
     fi
     out=$(basename $INPUT_HICPRO | sed -e 's/.allValidPairs.*/.mcool/')
 fi

conf/base.config

@@ -66,6 +66,11 @@ process {
     memory = { check_max( 10.GB * task.attempt, 'memory' ) }
     time = { check_max( 5.h * task.attempt, 'time' ) }
   }
+  withName:generate_cool {
+    cpus = check_max( 2, 'cpus' ) }                                                                                                                                                                        
+    memory = { check_max( 16.GB * task.attempt, 'memory' ) }
+    time = { check_max( 4.h * task.attempt, 'time' ) }   
+  }
 }
 
 params {